DUTCH OPERATIONAL SETUP

Toon Moene, 19th of May, 1998

Coming June the Dutch operational system will go from a 130x110 horizontal grid to 164x130, thereby covering all of the Mediterranean area. This change is mainly inspired by requests from the national television broadcaster, and has little meteorological value. Due to this change, we had to upgrade our (SGI) 4xR10000 PowerChallenge to an 8xR10000 Origin 2000. The vertical resolution remains the same (31 layers with the same A's and B's as the ECMWF operational model).

For the first time we tried to use rotated boundary fields (on a 1.5x1.5 degree resolution). We chose a slightly larger area than the forecast one, but in such a way that every point in the boundary field's grid is a point in the forecast grid.

As orography is not sent out with other fields in the operational dissemination from ECMWF, and one cannot obtain fields on rotated grids from MARS, we had to ask ECMWF staff to supply us with a matching orography. John Chambers worked hard to give us this field on the requested grid, and after a first failed attempt, the system is now functioning satisfactorily for over a month in pre-operational mode.

The timings of the new model configuration on the 8 processor Origin are as follows:

Analysis:

takes approximately 150 seconds

Forecast:

Version 2.7.15, using the Kuo/Cond option for convection/ condensation. 0.5x0.5 degree horizontal resolution, 3 minute timestep for dynamics and vertical diffusion, 12 minutes for physics. (In Oslo I reported 3500 seconds for this, but in the mean time I found out that compiling _all_ routines in grdy with automatic parallellisation on gives the best results).